Adaptive operation of large hydro-wind-solar integrated systems by aggregation-decomposition under inconsistent resource conditions

Development of large Hydro-Wind-Solar (HWS) power generation system involves complex hydraulic-electric linkages and severe resource variations. Inconsistent resources, reflected by changed statistical characteristics, may impair the system performance using traditional static operation rules extracted from historical series. Here, a framework for generating dynamic adaptive operation rules were proposed to deal with inconsistent resource conditions. First, a multivariate stochastic simulation model was used to generate inconsistent HWS scenarios. Second, a multiobjective optimization model was constructed, and static operation functions were extracted by aggregation-decomposition. Finally, adaptive operation rules were identified with a coupled “ensemble Kalman filter–k-nearest neighbors” algorithm and their parameters dynamically adjusted according to resource variations. This framework was applied to a clean-energy base at upper Yellow River, yielding the following results: (1) model simulation accuracy exceeded 99 %, and inconsistent scenarios were generated with mean values, Cv, seasonal variations; (2) independent and dependent variables for aggregation system were total input energy and total power output; dependent variables for two decomposition systems were release and carryover storage, identified from six independent variables; (3) under dynamic rules, total power generation and guaranteed rate were higher than under static rules by 5–6 % and 8–11 % respectively, and only mean variation affected operation benefits. Large benefit improvement and stable parameters appeared in downstream system.